Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Corollary 9·1 — (w_t − κ₁n_t)/(σ(t/μ₁)^{1/2}) and, if μ₂ < ∞, (w_t − (κ₁/μ₁)t)/(γt/μ₁)^{1/2} are asymptotically standard normal

Proved
SmithRegenerative.CLT.corollary_9_1

by mikedeng1 · Oct 6, 2026 · Mathlib 0df444a (Lean v4.33.1)

central-limit-theoremcumulative-processp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1regenerative-processrenewal-reward

Let wtw_twt​ be a cumulative process on a renewal process t1,t2,…t_1, t_2, \dotst1​,t2​,… with t0=0t_0 = 0t0​=0 and w0=0w_0 = 0w0​=0, with cycle increments yn=Δnwty_n = \Delta_n w_tyn​=Δn​wt​, variation increments y~n\tilde y_ny~​n​, and ntn_tnt​ the number of regenerations in [0,t][0,t][0,t]. Suppose E(Δnw~t)2<∞E(\Delta_n \tilde w_t)^2 < \inftyE(Δn​w~t​)2<∞ and μ1=Et1<∞\mu_1 = E t_1 < \inftyμ1​=Et1​<∞, and put

κ1=EΔnwt,κ2=E(Δnwt)2,σ2=κ2−κ12,σ>0.\kappa_1 = E \Delta_n w_t, \qquad \kappa_2 = E(\Delta_n w_t)^2, \qquad \sigma^2 = \kappa_2 - \kappa_1^2, \quad \sigma > 0.κ1​=EΔn​wt​,κ2​=E(Δn​wt​)2,σ2=κ2​−κ12​,σ>0.

Then:

  1. for every real α\alphaα,
lim⁡t→∞P{wt−κ1ntσ(t/μ1)1/2≤α}=Φ(α);\lim_{t \to \infty} P\left\{ \frac{w_t - \kappa_1 n_t}{\sigma (t/\mu_1)^{1/2}} \le \alpha \right\} = \Phi(\alpha);t→∞lim​P{σ(t/μ1​)1/2wt​−κ1​nt​​≤α}=Φ(α);
  1. if in addition μ2=Et12<∞\mu_2 = E t_1^2 < \inftyμ2​=Et12​<∞, and γ=var⁡Δn(wt−κ1μ1−1t)=var⁡(y1−(κ1/μ1)t1)\gamma = \operatorname{var} \Delta_n(w_t - \kappa_1 \mu_1^{-1} t) = \operatorname{var}(y_1 - (\kappa_1/\mu_1) t_1)γ=varΔn​(wt​−κ1​μ1−1​t)=var(y1​−(κ1​/μ1​)t1​) is positive, then for every real α\alphaα,
lim⁡t→∞P{wt−(κ1/μ1)t(γt/μ1)1/2≤α}=Φ(α).\lim_{t \to \infty} P\left\{ \frac{w_t - (\kappa_1/\mu_1) t}{(\gamma t/\mu_1)^{1/2}} \le \alpha \right\} = \Phi(\alpha).t→∞lim​P{(γt/μ1​)1/2wt​−(κ1​/μ1​)t​≤α}=Φ(α).

The second assertion is the central limit theorem for renewal-reward processes with a deterministic centring; applied to wt=ntw_t = n_twt​=nt​ it gives Feller's central limit theorem for renewal processes.

Formalization Note σ>0\sigma > 0σ>0 and γ>0\gamma > 0γ>0 are implicit in the paper, which divides by σ\sigmaσ and γ\sqrt\gammaγ​. The two assertions are the two conjuncts of the statement.

Preamble
import Mathlib
import Definitions.Def_SmithRegenerative_CLT_CumulativeProcess

open MeasureTheory ProbabilityTheory Filter Topology
Formal statement
namespace SmithRegenerative.CLT

/-- Smith (1955), Corollary 9·1, p. 30: "If `w_t` is a general cumulative process for which
`E(Δ_n w̃_t)² < ∞`, `κ₁ ≡ EΔ_n w_t`, `κ₂ ≡ E(Δ_n w_t)² < ∞`, and `σ² ≡ κ₂ − κ₁²`, then provided
`μ₁ < ∞`, `lim_{t=∞} P{(w_t − κ₁n_t)/(σ(t/μ₁)^{1/2}) ≤ α} = Φ(α)` (5·4·8). If, in addition,
`μ₂ < ∞`, then `lim_{t=∞} P{(w_t − (κ₁/μ₁)t)/((γt/μ₁)^{1/2}) ≤ α} = Φ(α)` (5·4·9), where
`γ = var Δ_n(w_t − κ₁μ₁⁻¹t)`."

For a cumulative process with `t₀ = 0`, `w₀ = 0` (the model with `M = 1`); the two assertions
are the two conjuncts, the second under `μ₂ = E t₁² < ∞`.

**Formalization Note** `σ > 0` and `γ > 0` are implicit (the paper divides by `σ` and `√γ`).
`Δ_n(w_t − κ₁μ₁⁻¹t) = y_n − (κ₁/μ₁) t_n`, so `γ = var(y₁ − (κ₁/μ₁) t₁)`. `n_t` is `count`.
`Φ = cdf (gaussianReal 0 1)`, and each limit is asserted for every real `α`. -/
theorem corollary_9_1 {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω) [IsProbabilityMeasure P]
    (τ : ℕ → Ω → ℝ) (w : ℝ → Ω → ℝ) (hw : IsCumulativeModel P τ (fun _ : Fin 1 => w))
    (hμ : Integrable (τ 1) P)
    (hvar2 : Integrable (fun ω => varIncr w τ 1 ω ^ 2) P)
    (σ : ℝ) (hσ_pos : 0 < σ)
    (hσ : σ ^ 2 = (∫ ω, incr w τ 1 ω ^ 2 ∂P) - (∫ ω, incr w τ 1 ω ∂P) ^ 2) :
    (∀ α : ℝ, Tendsto
      (fun t : ℝ => P.real {ω | (w t ω - (∫ ω', incr w τ 1 ω' ∂P) * (count τ t ω : ℝ)) /
          (σ * Real.sqrt (t / mu1 P τ)) ≤ α})
      atTop (𝓝 (cdf (gaussianReal 0 1) α))) ∧
    (Integrable (fun ω => τ 1 ω ^ 2) P →
      ∀ γ : ℝ, 0 < γ →
        γ = variance (fun ω => incr w τ 1 ω - (∫ ω', incr w τ 1 ω' ∂P) / mu1 P τ * τ 1 ω) P →
        ∀ α : ℝ, Tendsto
          (fun t : ℝ => P.real {ω | (w t ω - (∫ ω', incr w τ 1 ω' ∂P) / mu1 P τ * t) /
              Real.sqrt (γ * t / mu1 P τ) ≤ α})
          atTop (𝓝 (cdf (gaussianReal 0 1) α))) := by sorry

end SmithRegenerative.CLT
Source
Smith, Regenerative stochastic processes, Proc. R. Soc. Lond. A 232(1188):6–31 (1955), DOI 10.1098/rspa.1955.0198, p. 30, Corollary 9·1 (5·4·8)–(5·4·9)
Human review
  • Endorsed by Shuze Chen · Oct 6, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 6, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me